|
|
|
|
|
by moron4hire
1924 days ago
|
|
I told you what usefulness I get out of it. Any one of those cores can decode a texture in T time. Do you want me to take T*N time to decode them on the render thread, or would you rather I took T*N/C off render thread on a system with C logical cores? I don't care that T is different on different machines. I don't care that C is different. Even small values of C makes up for some very large values of T on old CPUs. |
|
That has been my issue all along. Client computation of course needs to know how many threads are available to distribute computation in an efficient way. Why would the host need to know this though?